R: Compute logical OR over subset of columns with matches

Question

I have this dataframe:

df <- tibble(id = c(1, 2, 3), c_1 = c(T, T, F), c_2 = c(F, F, T)) %>% group_by(id)
# A tibble: 3 x 3
     id c_1   c_2  
  <dbl> <lgl> <lgl>
1     1 TRUE  FALSE
2     2 TRUE  FALSE
3     3 FALSE TRUE 

I now want to compute the rowise logical OR over the columns starting with c_ I tried

df %>% mutate(valid = sum(select(matches("^c_")) == 0))

However I get

`matches()` must be used within a *selecting* function.

how can i solve this?

Answer 1

library(dplyr)

df <- tibble(id = c(1, 2, 3), c_1 = c(T, T, F), c_2 = c(F, F, T))

df %>% 
  rowwise() %>% 
  mutate(
    valid = any(c_across(starts_with("c_")))
  ) %>% 
  ungroup()

#> # A tibble: 3 × 4
#>      id c_1   c_2   valid
#>   <dbl> <lgl> <lgl> <lgl>
#> 1     1 TRUE  FALSE TRUE 
#> 2     2 TRUE  FALSE TRUE 
#> 3     3 FALSE TRUE  TRUE

^{Created on 2022-07-11 by the reprex package (v2.0.1)}

Answer 2

Update: Why is tibble() needed:

Without as_tibble() or tibble() or data.frame() it won't work:

Your table:

> class(df)
[1] "grouped_df" "tbl_df"     "tbl"        "data.frame"

• Without as_tibble() or tibble() or data.frame() -> won't work:

>df %>%
  mutate(valid = ifelse(rowSums(select(., contains("c_")))==1, TRUE, FALSE))

Adding missing grouping variables: `id`
Error in `mutate()`:
! Problem while computing `valid = ifelse(rowSums(select(.,
  contains("c_"))) == 1, TRUE, FALSE)`.
x `valid` must be size 1, not 3.
i The error occurred in group 1: id = 1.

• With as_tibble() or tibble() or data.frame() -> it will work:

df %>% 
  data.frame() %>% 
  mutate(valid = ifelse(rowSums(select(., contains("c_")))==1, TRUE, FALSE))

#or

df %>% 
  tibble() %>% 
  mutate(valid = ifelse(rowSums(select(., contains("c_")))==1, TRUE, FALSE))

First answer: If we want to do it with select: Here is an out of the box approach:

library(tibble)
library(dplyr)

df %>% 
  as_tibble() %>% 
  mutate(valid = ifelse(rowSums(.[2:3])==1, TRUE, FALSE))

or

library(tibble)
library(dplyr)

df %>% 
  as_tibble() %>% 
  mutate(valid = ifelse(rowSums(select(., contains("c_")))==1, TRUE, FALSE))

# A tibble: 3 x 4
     id c_1   c_2   valid
  <dbl> <lgl> <lgl> <lgl>
1     1 TRUE  FALSE TRUE 
2     2 TRUE  FALSE TRUE 
3     3 FALSE TRUE  TRUE 

Answer 3

We may use if_any directly without rowwise

library(dplyr)
df %>%
   mutate(valid = if_any(starts_with('c_')))
# A tibble: 3 × 4
     id c_1   c_2   valid
  <dbl> <lgl> <lgl> <lgl>
1     1 TRUE  FALSE TRUE 
2     2 TRUE  FALSE TRUE 
3     3 FALSE TRUE  TRUE 

data

df <- tibble(id = c(1, 2, 3), c_1 = c(TRUE, TRUE, FALSE),
 c_2 = c(FALSE, FALSE, TRUE))

Answer 4

base R option using grepl:

library(dplyr) # For tibble
df <- tibble(id = c(1, 2, 3), c_1 = c(T, T, F), c_2 = c(F, F, T)) %>% group_by(id)
df$valid <- apply(df, 1, function(x) any(x %in% grepl("c_" , names(x))))
df
#> # A tibble: 3 × 4
#> # Groups:   id [3]
#>      id c_1   c_2   valid
#>   <dbl> <lgl> <lgl> <lgl>
#> 1     1 TRUE  FALSE TRUE 
#> 2     2 TRUE  FALSE TRUE 
#> 3     3 FALSE TRUE  TRUE

^{Created on 2022-07-11 by the reprex package (v2.0.1)}